> For the complete documentation index, see [llms.txt](https://docs.trueopen.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.trueopen.ai/protocol/design-principles.md). # Design Principles The reasoning behind TrueOpen's execution, verification, and accountability boundaries. TrueOpen coordinates inference across independent participants. These principles explain the design; the [V1 rules](/protocol/v1.md) specify the behavior that implements it. ## Off-chain execution, on-chain accountability Model execution and bulk Task data stay off-chain. The chain records commitments, assignments, deadlines, and accounting. This keeps the consensus workload separate from model computation while giving participants a shared record of responsibility. ## Independent verification Execution and judgment are distinct duties. A Worker produces output and evidence; independently assigned Verifiers evaluate that material using the frozen Profile. Commit-reveal separates committing to a result from disclosing it. Challenges provide an additional independent round under the protocol's rules. Verification is bounded by the selected Profile's algorithm and evidence requirements. It does not establish that every generated statement is factually true. ## Freeze inputs before selecting participants Candidate facts freeze before the selection beacon becomes known. Future randomness limits the ability to tailor a candidate set to a known draw. Domain separation keeps different selection purposes independent. See [Selection and randomness](/protocol/v1/randomness.md). ## Make responsibility explicit An offer of capacity, a transport acknowledgement, and an accepted assignment are different facts. The chain determines when a provider becomes accountable. ServiceBond liability belongs to the service operator, while consensus stake protects a separate responsibility domain. ## Preserve historical meaning Tasks retain their accepted Profile, order, rule versions, and parameter snapshots. Later changes must not reinterpret work already accepted. This allows independent participants to verify the same result even when software and network settings evolve. ## Settle deterministically Success, timeout, unavailable data, and failed verification need defined outcomes. Funds remain separated by purpose, and settlement records finality together with the accounting result. Missing facts or violated invariants must not be repaired with local guesses. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.trueopen.ai/protocol/design-principles.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.